5 ms·

Dunno about contemporary music players, but: > no tune was played twice until all of the tunes were played That is definitely not compatible with actual mathe

Dunno about contemporary music players, but: > no tune was played twice until all of the tunes were played That is definitely not compatible with actual mathematical “randomness”. Try rolling a die multiple times and see whether you get all six numbers before repeats. It’s very unlikely (probability of all six rolls being unique = 6!/6^6 = 1.5%) Which brings up the interesting question — what do humans actually expect from “shuffle” mode?

larrik 7y ago
I don't think mathematics is the issue here, rather problem definition. OP wants "Shuffle my playlist into a random order and then play it through" You are thinking closer to "After this song is done, choose a new song at random among ALL available." Spotify's is a mixture, while good offline clients tend to go to the first.
siphon22 7y ago
Personally what i want out of shuffle is that once a song has been replayed more than n amount of times, it gets temporily removed from the queue for that session or an amount of time.
simion314 7y ago
In statistics classes we learned that all problems can be reduced to problems with balls and urns. In this problems you can put the ball back in the urn or not put it back, both are valid things and can model different problems.
egypturnash 7y ago
Think of every single erroneous belief that people have about randomness. A pleasing music-randomizer algorithm obeys all of these. The term found on a lot of music players is not "random", it is "shuffle", and this is an important distinction. The term comes from cards. Take a deck of cards, shuffle it, and turn each card over in order from top to bottom. You expect that every card in the deck will be seen once and only once; turning up, say, two eights of clubs in a row means that something is dreadfully wrong with your deck. Humans tend to expect a similar behavior from "shuffle" modes: the same song does not repeat for a while. If people are shuffling by song, then they may also expect that performers don't repeat within a (smaller) timeframe. Nobody expects "pick a random number between 1 and numsongs, play that song, repeat, and be perfectly happy to play Baby Shark forty-seven times in a row if that number keeps on coming up".
mkl 7y ago
Mathematical randomness can describe an enormous range of behaviour, including shuffling. Shuffling is sampling without replacement. Rolling a die repeatedly is sampling with replacement (imagine the numbers 1-6 written on cards and mixed together in a bag; to get a die's behaviour you need to put a card you draw out and look at back again before the next draw).
ska 7y ago
> That is definitely not compatible with actual mathematical “randomness”. Of course it is, the concept of randomness is much broader than you allow for. The specific issue you refer to here is if you are sampling with or without replacement - both are valid. There is another issue which is the distribution you use. Should songs be equally likely to occur, or should the ones you've played more occur more likely? You may also extend sampling strategies to genre, or to exclude repeated artists, or whatever - but all of these things could still be done via random sampling.

larrik 7y ago

I don't think mathematics is the issue here, rather problem definition. OP wants "Shuffle my playlist into a random order and then play it through" You are thinking closer to "After this song is done, choose a new song at random among ALL available." Spotify's is a mixture, while good offline clients tend to go to the first.

siphon22 7y ago

Personally what i want out of shuffle is that once a song has been replayed more than n amount of times, it gets temporily removed from the queue for that session or an amount of time.

simion314 7y ago

In statistics classes we learned that all problems can be reduced to problems with balls and urns. In this problems you can put the ball back in the urn or not put it back, both are valid things and can model different problems.

egypturnash 7y ago

Think of every single erroneous belief that people have about randomness. A pleasing music-randomizer algorithm obeys all of these. The term found on a lot of music players is not "random", it is "shuffle", and this is an important distinction. The term comes from cards. Take a deck of cards, shuffle it, and turn each card over in order from top to bottom. You expect that every card in the deck will be seen once and only once; turning up, say, two eights of clubs in a row means that something is dreadfully wrong with your deck. Humans tend to expect a similar behavior from "shuffle" modes: the same song does not repeat for a while. If people are shuffling by song, then they may also expect that performers don't repeat within a (smaller) timeframe. Nobody expects "pick a random number between 1 and numsongs, play that song, repeat, and be perfectly happy to play Baby Shark forty-seven times in a row if that number keeps on coming up".

mkl 7y ago

Mathematical randomness can describe an enormous range of behaviour, including shuffling. Shuffling is sampling without replacement. Rolling a die repeatedly is sampling with replacement (imagine the numbers 1-6 written on cards and mixed together in a bag; to get a die's behaviour you need to put a card you draw out and look at back again before the next draw).

ska 7y ago

> That is definitely not compatible with actual mathematical “randomness”. Of course it is, the concept of randomness is much broader than you allow for. The specific issue you refer to here is if you are sampling with or without replacement - both are valid. There is another issue which is the distribution you use. Should songs be equally likely to occur, or should the ones you've played more occur more likely? You may also extend sampling strategies to genre, or to exclude repeated artists, or whatever - but all of these things could still be done via random sampling.